Nobody reads your container logs. Dozzle Cloud does.
Free turns the alerts you already have into one message that says what broke. Pro reads your logs every morning, across every host, and finds the problems nothing alerted on — out of the 75 distinct errors a day the average stack logs.
Two ways to hear about it.
Alerting tells you about the rules you thought to write. The daily review catches what nobody thought to watch for. The first is the free tier; the second is what Pro adds.
Notifications and triage
FreeThe moment it breaks
Write alert rules in your own Dozzle for the containers you care about. Dozzle Cloud triages what fires, groups the repeats, and sends one alert instead of forty — with quiet hours and severity filters on every channel.
Proactive log analysis
ProEvery morning
Reads the logs you are already sending, compares CPU, memory and restarts with last week, and reports what changed — including problems no rule was ever written for. Every host you run, not just one. Not real time, and not trying to be. Free gets the most serious finding once a week.
Ask the agent
When you want the why, just ask.
Every alert is one tap from a conversation. Walk through the failure, pull related logs, see what changed. In Telegram, Discord, or the web app.
Or bring your own AI
PreviewMCP for Claude, Cursor, anything.
Wire Dozzle Cloud into the assistant you already use. Read-only access to events, logs, and container state. Free on every plan.
How it works--url https://mcp.dozzle.dev/mcp
--header "Authorization: Bearer dz_mcp_…"
Proactive log analysis
ProIt goes looking. You don't have to set anything up.
- 1
It reads your logs, stats and restarts
Every morning on Pro, with no alert rules to write. Error logs, CPU and memory against last week, every restart and redeploy — so a day with clean logs still gets looked at.
- 2
Groups them into real problems
Three hundred lines of the same error are one problem, not three hundred. The typical stack has 75 of them.
- 3
Investigates what looks serious
Pulls logs, inspects the container, checks what changed — and weighs what it already told you yesterday.
- 4
It decides what to do
With all of that context, every event lands in one of three places.
- Ignore
- Seventy-two of the seventy-five. Normal for your stack, and it stays that way.
- Track
- Still happening on day four, three times worse than yesterday. You see that it changed, not that it exists.
- Tell you
- Two or three things worth your morning, with what broke and where.
Notifications and triage
FreeOne alert. The whole story.
Cause first. Counts grouped. Sent wherever you already are.
Indexed on arrival
FreeSearch inside Dozzle, stats you keep.
Every event from every connected Dozzle is searchable the moment it arrives — across hosts, from the browser, without opening a terminal. CPU, memory, network and disk are stored as history, so a spike at 3am is still there in the morning. Neither counts against your event limit.
Any event, any host, any time
The spike you slept through
03:14 cpu 91.4% ██████████████████
03:14 memory 62.0% ████████████
07:02 cpu 8.1% ██
07:02 memory 31.4% ██████“I want to know when my containers crash overnight. But I don't want to get spammed about every incident over and over. One notification is enough.”
Free keeps you quiet. Pro goes looking.
Free tier, no credit card. Pro is $5/mo — see pricing.
